Matrix Completion with Incomplete Side Information via Orthogonal Complement Projection

Authors: Gengshuo Chang, Wei Zhang, Lehan Zhang

ICML 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results show that the proposed OCMC model outperforms state-of-the-art methods on both synthetic data and real-world applications. ... In this section, we demonstrate the effectiveness of the proposed OCMC model and the linear ADMM algorithm through experiments conducted on both synthetic experiments and real-world applications. In the synthetic experiments, we primarily investigated the relationship between matrix completion accuracy and the number of observations under varying completeness levels of side information. In the real-world application experiments, we focused on the multi-label learning (Goldberg et al., 2010) and movie recommendation (Harper & Konstan, 2015).
Researcher Affiliation Academia 1Department of Information and Communication Engineering, Harbin Institute of Technology, Shenzhen. Correspondence to: Wei Zhang <EMAIL>.
Pseudocode Yes Algorithm 1 Linearized ADMM for OCMC with Squared Loss
Open Source Code No The paper does not explicitly state that source code for the described methodology is being released or provide a direct link to a code repository. It mentions
Open Datasets Yes For real-world datasets, including movie recommendation and multi-label learning tasks, results show that OCMC consistently outperforms state-of-the-art methods... In this experiment, we use the Movie Lens-100k dataset (Harper & Konstan, 2015)... The dataset we selected is the web page classification from Yahoo.com (Ueda & Saito, 2002).
Dataset Splits Yes We randomly pick 10% instances as the test set and use the rest 90% as the training set. ... The data is divided into a training set and a test set, with the proportion of the test set ranging from 0.1 to 0.9 in increments of 0.2.
Hardware Specification No The paper does not provide specific hardware details (e.g., GPU/CPU models, memory) used for running its experiments.
Software Dependencies No The paper does not provide specific ancillary software details (e.g., library or solver names with version numbers) needed to replicate the experiment.
Experiment Setup No The paper lists parameters for Algorithm 1 (λ, β0, βmax, ρ, τ) but does not provide their specific values used in the experiments. It describes general experimental conditions such as sampling rates and completeness levels, but not concrete hyperparameter values or detailed training configurations.